Get Complete Project Material File(s) Now! »
Bioreactor heterogeneities and bioperformances
During cell cultures in bioreactors, interactions between physico-chemical parameters (microorganism environment) and biological systems (bio-reactivity) affect bioprocess performances. These phenomena were previously described and explained in several handbooks (Asenjo and Merchuk 1994; Kent 2012; Soetaert and Vandamme 2010). The scientific and technological continuum relies on three major contributions: bioreactor design, transfer control and microorganism metabolism. Consequently, bioperformances result from the control and the understanding of both physical and biological responses at global and local scales (Figure I-1).
Origin of heterogeneities inside bioreactor
Bench-scale bioreactors with cultivation volumes typically ranging from 0.2 to 20 L are generally considered as perfectly mixed. Such bioreactors are usually equipped with adequate mixing systems that ensure a good homogeneity of the fermentation broth and therefore high mass and heat transfer capabilities. When scaling up a bioprocess to an industrial production scale, the increase in size of bioreactors is often associated with a poor mixing efficiency, which leads to mass and heat transfer limitations and consequently causing inhomogeneities within large volume cultures (Carlquist et al. 2012; Delvigne and Goffin 2014; Enfors et al. 2001; Hewitt et al. 2007; Lara et al. 2006; Muller et al. 2010; Palomares and Ramirez 2000).
Nature of heterogeneities in industrial bioreactors
Heterogeneities in industrial bioreactors, as absolute values or generated gradients, have been identified and documented in several handbooks (Asenjo and Merchuk 1994; Kent 2012; Soetaert and Vandamme 2010) and scientific papers (Amanullah et al. 2001; Baert et al. 2016; Bylund et al. 1998; Carlquist et al. 2012; Enfors et al. 2001; Fernandes et al. 2012; Kuschel et al. 2017; Langheinrich and Nienow 1999; Lara et al. 2006; Larsson et al. 1996; Manfredini et al. 1983; Nagy et al. 1995; O’Beirne and Hamer 2000; Phue and Shiloach 2005; Schweder et al. 1999; Tsao et al. 1992; Varma and Palsson 1994; Wick et al. 2001; Xu et al. 1999). It affects fundamental culture parameters such as shear rate, dissolved oxygen concentrations, substrate concentrations, pH, temperature and osmotic pressure.
Cells are transiently exposed to different shear forces as they circulate throughout the various zones of a bioreactor. Duration and intensity of such forces depend on the impeller type, agitation, aeration rate, and reactor configuration. Several deleterious effects of shear have been observed in cultured cells. Animal and fungal cells are especially the most sensitive to mechanical stress (Chisti 2001; Cruz et al. 1998; Manfredini et al. 1983; Palomares et al. 2006). Plant cells and microbial cultures may also be influenced by the effects of shear (Sahoo et al. 2006; Taticek et al. 1991; Toma et al. 1991).
Shear rates of 1482 s-1 have been observed to cause loss in viability in B. subtilis cell cultures (Sahoo et al. 2006). It may trigger several physiological responses, which lower biomass and product yield or affect product quality (Lu et al. 1995; McDowell and Papoutsakis 1998; Senger and Karim 2003). Besides, shearing flow can induce transcription factors (Ranjan et al. 1996) and can influence DNA synthesis rate and the proliferative state of cells (Lakhotia et al. 1992). Nevertheless, in some cases, positive effects of shear have been observed within certain limits. In fact, it has been proved that shear may improve the Microcin B17 (antibiotic) production by Escherichia coli ZK650 strain (Gao et al. 2001). Such positive effects may be due to enhancement of heat and mass transfer rates (Asenjo and Merchuk 1994).
Dissolved oxygen concentration
The oxygen transfer rate is a critical parameter in bioreactors. Cells passing through the oxygen-limited zones sense and respond to oxygen depletion, leading to the activation of microaerobic or anaerobic metabolism and therefore to the accumulation of by-products.
Experiments carried out by Larsson et al. (1996) demonstrated that oxygen gradients vary with stirrer speed, the time of cultivation and the organism type and process. Oosterhuis and Kossen (1984) identified oxygen gradients ranging from 22 % (with respect to air saturation) in well-mixed zones to 0 % in stagnant regions of a 19 m3 bioreactor. In 1994, Reuss and co-workers (1994) employed computational fluid dynamics (CFD) techniques to simulate oxygen distribution within a 100 L bioreactor. They predicted a 64-fold difference between maximum and minimum oxygen concentrations within the reactor. Similarly, Zahradnik and colleagues (2001) developed a model based on the Networks-of-Zones approach for the description of gas-liquid flow in a 3 m3 triple-impeller stirred bioreactor. The Networks-of-Zones approach is based on a spatial subdivision of the reactor into finite volumes. These computations are more tractable than full computational fluid dynamics (CFD) and provide a realistic picture of the internal behavior of a reactor. Results showed spatial non-uniformities in the gas holdup varying from 2 to 80 % (with a mean value of 17 ± 0.9 %) (Figure I-2A) and local dissolved oxygen concentrations ranging from 3 to 49 % (with a mean value of 35 ± 0.24 %) (Figure I-2B). Under considered operating conditions, this simulation shows the importance of dissolved oxygen variations in different zones of a 3 m3 bioreactor.
Further experiments based on the simulations of gas-liquid flow, using CFD techniques coupled with population balance equations have been run in a 20 L multi-impeller bioreactor (equipped with six-bladed Rushton impellers located at the center of the vessel) (Zhang et al. 2009). Population balance is a well-established method used to analyze the size distribution of the dispersed phase and accounting for breakage and coalescence effects. An inhomogeneous distribution of gas bubble size, gas holdup and turbulent energy dissipation has been predicted throughout the whole bioreactor. These perturbations are thought to be very important in determining the health and viability of microorganisms.
Moreover, experiments using a dual probe technique demonstrated the existence of dissolved oxygen (DO) gradients in a 5000-L bioreactor (Xing et al. 2009). Two DO probes, one placed at the bottom of the bioreactor and the other was located just below the liquid phase on the opposite side, were utilized to measure differences in the DO readings between the top and bottom of 20- and 5000-L bioreactors. Each bioreactor was equipped with three impellers, and a sparger containing holes of 15 µm (20-L bioreactor) and 1.8 mm (5000-L bioreactor) diameters. As shown in Figure I-3A, the top and bottom DO profiles were approximately identical in the 20-L vessel. Nevertheless, DO readings were different between both probe locations in the 5000-L bioreactor (Figure I-3B). Besides, the bottom DO probe responds more rapidly than the top probe that did not respond for over 6 min. These gradients were thought to be attributed to the lower mixing efficiency at the large scale.
pH is an important parameter that must be carefully measured and controlled during bioprocessing. Spatial pH fluctuations can occur in large-scale reactors near the base or acid addition points as a consequence of a deficient mixing. In addition, accumulation of by-products (e.g. organic acids) can generate low pH zones when cells are exposed to high substrate concentrations (overflow metabolism) or oxygen-limited conditions (micro/anaerobic metabolism). pH fluctuations may affect product quality, culture viability and biological functions (Amanullah et al. 2001; Onyeaka et al. 2003).
Structured models for predicting pH profiles in a 100 L bioreactor have been designed by Reuss et al. (1994). They predicted pH variations from 4.0 in the bulk liquid to 9.0 in the alkali addition zones. Langheinrich and Nienow (1999) found pH gradients of up to 1 unit between the base-feeding zone (liquid surface) and the bulk liquid in an industrial scale 8 m3 stirred bioreactor fitted with a Rushton turbine. Two pH probes were used to detect pH variations: one situated at the liquid surface (position 1) and the second in the impeller plane (position 3). They also demonstrated that changing the point of alkali addition from the liquid surface (Figure I-4A) to the well-mixed impeller regions (Figure I-4B) can reduce pH heterogeneities inside bioreactor.
A Similar approach based on dual pH probes measurements was used to investigate the occurrence of pH gradients in a 5000-L bioreactor (Xing et al. 2009). The two pH sensors were placed at distinct position in the vessel: at the bottom of the bioreactor and at the liquid surface (on the opposite side). The obtained results showed differences in the pH probes responses when adding the tracer at the liquid surface. Indeed, the top probe response was characterized by a rapid peak (pH overshoot) followed by a steady state decline until the final pH value, whereas the bottom sensor response showed a steady state increase up to a the same final pH. The pH overshot in the 5000-L bioreactor was approximately 1.67 units. It was reduced when decreasing the bioreactor volume up to 0.08 and 0.81 units for 5- and 20-L reactors, respectively.
A recent study (Cortes et al. 2016) provided a simulation of the pH gradients during batch cultures of an engineered Escherichia coli strain via a two-compartment scale-down system.
The system was composed of two interconnected stirred tank reactors (STR): one representing the conditions of the bulk of the fluid (well-mixed region) (STR1), and the second corresponding to alkali addition zone for pH regulation (STR2). A continuous circulation of cells between the two vessels was ensured by means of an external peristaltic pump. pH values were monitored at various residence times (tR) of 60, 120, 180 and 240 s. The residence times were simulated by setting a recirculation flow between the vessels. Figure I-5 represents the simulation of pH gradients at each residence time during the time course of fermentation. The pH gradient was present through almost the entire duration of the culture.
Values of pH gradients up to 2 units were attained in the alkali addition region at a tR = 240 s.
The largest pH gradient observed at a tR = 60 s was around 0.2 units.
Many industrial bioprocesses usually use fed-batch mode where high concentration of feed substrate is added at one point on the top of bioreactor leading to spatial substrate gradients within the culture broth.
Bylund and co-workers (1998) have predicted an inhomogeneous distribution of glucose concentrations in a 12 m3 fed-batch bioreactor, equipped with three Rushton turbines. The feed solution was highly concentrated (552 g L-1) in order to avoid dilution of bioreactor. The results demonstrated a gradual decrease of glucose concentrations away from the feed point, and concentrations of up to 400-fold the mean value were found at the level closest to the addition zone: concentrations changed from 5 mg L-1 most distant from the feed point up to almost 2000 mg L-1 close to the substrate addition level.
Similarly, Enfors et al. (2001) have predicted by CFD calculations the existence of glucose gradients when a 500 g L-1 glucose solution was added at the top of a 22 m3 fed-batch bioreactor, fitted with four Rushton turbines (Figure I-6). Substrate concentrations varying from around 0 to 3 g L-1 have been expected in the vicinity of the feed point (Figure I-7A). Whereas, glucose fluctuations (almost from 0 to 20 mg L-1 (Pos 2) and from 50 up to 250 mg L-1 (Pos 1)) have been observed in two points of the bulk liquid (distant from the feed point) (Figure I-7B).
Temperature gradients are not typically expected in large scale bioreactors (Manfredini et al. 1983). This parameter is routinely well controlled and maintained constant during cell culture. However, under some conditions (e.g. inefficient cooling systems, exothermic reactions) cells could be exposed to temperature fluctuations which may represent an environmental stress affecting their viability and activity. Specifically, Tsao et al. (1992) have observed that heat shock (from 37 to 42 °C during 1 h) induces different stress responses in mammalian CHO cell cultures, such as reduction of heterologous protein production, decrease of RNA translation rates, and in extreme cases, cell death. Furthermore, Gorenflo et al. (2007) have predicted the existence of temperature gradients in a 10 L culture of CHO cells. They observed a decrease in growth rate when cells were repeatedly exposed to 1 min excursions from 37 to 27 °C.
Impact on bioprocess performances
Local environmental gradients are typically encountered in many industrial scale fermentation processes. Cultivated microorganisms are then forced to experience sudden environmental changes as they circulate from one zone to the other. These perturbations may pose different types of stress (e.g. oxidative, temperature, pH) on the cells and affect their metabolism and physiology (Amanullah et al. 2001; Bylund et al. 1998; Carlquist et al. 2012; Delvigne et al. 2009; Enfors et al. 2001; Kuschel et al. 2017; Takors 2012). Nature, intensity, duration and/or frequency of these stress conditions would be responsible for different metabolic and physiological cell behaviors within the culture broth. The resulted various types of cell responses are expected to cause performance variability in terms of yield, titer and/or productivity in industrial bioreactors comparing to small-scale bioreactors (Enfors et al. 2001; Lara et al. 2006; Muller et al. 2010). Identification and quantification of the impacts of these heterogeneities on microbial dynamics is therefore necessary to achieve optimal performances at the industrial level.
The microorganism of interest in this research “Yarrowia lipolytica” is an ascomycete yeast with biotechnological potential to produce notably specific lipids for bio-jet fuels synthesis (Cescut 2009; Portelli 2011). Under partially controlled conditions (in presence of stress), the fungus is characterized by its ability:
ü to mediate metabolic shifts (e.g. production of organic acids) (Beopoulos et al. 2009), which may affect the conversion yields of substrate into biomass (YX/S) and/or lipids (YLip/S) causing thereby a decrease in the bioprocess performances.
ü to undergo a dimorphic transition from yeast to the mycelium forms in response to environmental stressors (Barth and Gaillardin 1997; Bellou et al. 2014; Braga et al. 2016; Kim et al. 2000; Palande et al. 2014; Ruiz-Herrera and Sentandreu 2002). The presence of mycelial cells affects significantly the rheological behavior of the broths (Kraiem et al. 2013) and the transfer phenomena in bioreactors (Kar et al. 2011; O’Shea and Walsh 2000) leading consequently to the deterioration of the cell productivity (Fickers et al. 2009; Galvagno et al. 2011).
To ensure a successful scale-up of Y. lipolytica-based processes from laboratory to production scale, it is of utmost importance to characterize the dynamics of its responses to perturbations in the growth environment, at both metabolic and morphological levels.
Physiological responses of Yarrowia lipolytica to stress conditions
General description of Yarrowia lipolytica
Yarrowia lipolytica (formerly known as Candida, Endomycopsis or Saccharomycopsis lipolytica) is a strictly aerobic yeast, belonging to the family of Hemiascomycetes (Barth and Gaillardin 1996; Barth and Gaillardin 1997). This yeast is usually isolated from dairy products such as cheese, yogurt, and sausages (Barth and Gaillardin 1996), from oily environments such as polluted soils, raw poultry or dairy products (Deak 2001; Lanciotti et al. 2005; Sinigaglia et al. 1994; Yalcin and Ucar 2009), as well as from marine and hypersaline media (Beopoulos et al. 2009).
Y. lipolytica is unable to grow above 32 °C. It is considered non-pathogenic and several processes based on this microorganism were classified as Generally Recognized As Safe (GRAS) by the Food and Drug Administration (FDA, USA) (Barth and Gaillardin 1996; Barth and Gaillardin 1997; Fickers et al. 2005).
Y. lipolytica is an industrially important microorganism capable to metabolize a wide variety of carbon substrates (e.g. glucose, alcohols, acetate and hydrophobic substrates) (Fickers et al. 2005; Finogenova et al. 2002; Papanikolaou et al. 2006), to degrade efficiently several low-value or toxic compounds (e.g. raw glycerol, olive mill wastewater) (Levinson et al. 2007;
Makri et al. 2010; Rymowicz et al. 2006; Sarris et al. 2011), and to produce a broad spectrum of valuable metabolites (e.g. organic acids, lipids, enzymes and proteins) (Bellou et al. 2016; Bussamara et al. 2010; Lopes et al. 2009a; Madzak et al. 2004; Papanikolaou and Aggelis 2009; Papanikolaou et al. 2007; Parfene et al. 2013; Ron and Rosenberg 2002; Sauer et al. 2008). These potentialities have increased interest in the exploitation of Y. lipolytica in numerous biotechnological and environmental applications (Coelho et al. 2010; Goncalves et al. 2014; Ledesma-Amaro and Nicaud 2016; Liu et al. 2015; Zinjarde 2014).
Table of contents :
INTRODUCTION AND CONTEXT OF THE STUDY
PART I: LITERATURE REVIEW
I.1 Bioreactor heterogeneities and bioperformances
I.1.1 Origin of heterogeneities inside bioreactor
I.1.2 Nature of heterogeneities in industrial bioreactors
I.1.3 Impact on bioprocess performances
I.2 Physiological responses of Yarrowia lipolytica to stress conditions
I.2.1 General description of Yarrowia lipolytica
I.2.2 Morphological changes in Yarrowia lipolytica: dimorphism phenomena
I.2.3 Metabolic responses of Yarrowia lipolytica to adverse environmental conditions
I.2.4 Mechanisms regulating stress responses in Yarrowia lipolytica
I.3 Conclusions and objectives of the study
PART II: MATERIALS AND METHODS
II.1 Strain and culture media
II.1.2 Culture media
II.2 Culture conditions
II.2.1 Strain preservation
II.2.2 Inoculum preparation
II.2.3 Bioreactor fermentation
II.3 Model particles
II.3.1 Calibration particles: from 1 to 15 μm-diameter
II.3.2 Calibration particles: 40 and 80 μm-diameter
II.4 Analytical methods
II.4.1 Biomass characterization
II.4.2 Determination of glucose and organic acids concentrations
II.4.3 Gas analysis
II.5 Methodology for data treatment
II.5.1 Rates expression for gas-phase reactions
II.5.2 Rate expressions for liquid-phase reactions
II.5.3 Determination of overall yields
II.5.4 Carbon and redox balances
II.5.5 Smoothing and reconciliation of the experimental data
PART III: RESULTS AND DISCUSSION
Chapter III-1: Development and validation of methodologies for the quantification and characterization of morphological changes in Yarrowia lipolytica
III.1.2 Experimental strategy
III.1.3 Results and discussion
III.1.5 Results synthesis
Chapter III-2: Characterization of the mycelial transition of Yarrowia lipolytica by morphogranulometric measurements
III.2.2 Experimental strategy
III.2.3 Results and discussion
III.2.5 Results synthesis
Chapter III-3: Dynamic behavior of Yarrowia lipolytica in response to pH perturbations: dependence of the stress response on the culture mode
III.3.2 Publication: Dynamic behavior of Yarrowia lipolytica in response to pH perturbations: dependence of the stress response on the culture mode
III.3.3 Results synthesis
Chapter III-4: Investigation of the effect of oxygen availability on the metabolism and morphology of Yarrowia lipolytica: more insights into the impact of glucose levels on dimorphism
III.4.2 Publication: Investigation of the effect of oxygen availability on the metabolism and morphology of Yarrowia lipolytica: more insights into the impact of glucose levels on dimorphism
III.4.3 Results synthesis
GENERAL DISCUSSION, CONCLUSIONS AND PERSPECTIVES